ML-based Approaches for Joint SAR Imaging and Phase Error Correction

This paper addresses a series of iterative sparse recovery approaches with application to the synthetic aperture radar (SAR) imaging which suffers from motion-induced model errors. These types of errors result in phase errors in SAR data, which cause defocusing of the reconstructed images. The proposed phase-error correction approaches combine the maximum a posterior (MAP) algorithm and the iterative sparse maximum likelihood-based (SMLA) approaches (referred to as the PE-MAP-SMLA approaches) to solve a joint optimization problem to achieve phase errors estimation and SAR image formation simultaneously. A new PESLIM approach is also proposed that extends the idea of the classical sparse and learning via iterative minimization (SLIM) approach. A closed-form expression for the recursive estimate of the phase errors parameters is derived. A general form of each of these iterative approaches consists of three steps, the first of which is for image formation, the second is for phase errors estimation and the last is for nuisance parameters estimation. The proposed approaches can correct the phase errors accurately, and the reconstruction quality of the SAR images can be improved significantly. Finally, simulation results of 1-D spectral estimation and 2-D SAR imaging examples are generated to show the effectiveness of the proposed approaches.